Repetitive Motion Detection Without Training Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing motion detecting devices, particularly those using machine learning approaches, are limited in their ability to detect repetitive motions across various types of applications due to the need for well-defined target motions and corresponding training datasets.
Innovation Solution
A method and device that receive and process accelerated and angular velocities to determine stationary points, motion periods, and cluster these periods based on differences in angular velocities, allowing for the identification of repetitive motions without requiring specific training datasets for each type of motion.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If machine learning approach is used for repetitive motion detection, then detection accuracy for specific motions is improved, but device complexity and requirement for training datasets increase
Solution Approach 1:
The patent extracts the essential characteristics of repetitive motions (stationary points, motion periods, angular velocity patterns) from the complex machine learning approach. By identifying and analyzing these key features directly from sensor data, the system achieves motion detection without requiring extensive training datasets, thus resolving the contradiction between detection accuracy and system complexity
Solution Approach 2:
The patent replaces the machine learning system (which requires training data) with a rule-based analysis system that processes angular velocity and acceleration data through defined algorithms. This substitution eliminates the need for training datasets while maintaining the ability to detect and classify repetitive motions through physical parameter analysis
2Measurement precision
If machine learning approach is used for repetitive motion detection, then detection accuracy for specific motions is improved, but adaptability to different applications is reduced
Solution Approach 1:
The patent creates a universal motion detection framework that can identify repetitive motions across different applications (sports training, rehabilitation, industrial monitoring) using the same core algorithms. By focusing on fundamental motion characteristics rather than application-specific patterns, the system achieves both accuracy and versatility, resolving the contradiction between specialized detection and broad adaptability
3Measurement precision
If clustering is performed on motion periods, then identification of repetitive patterns is improved, but processing time increases
Solution Approach 1:
The patent segments the continuous motion data into discrete motion periods based on stationary points and angular velocity thresholds. This segmentation allows the system to process and cluster only relevant motion segments rather than analyzing entire data streams, improving repetitive pattern identification while reducing overall processing time
Solution Approach 2:
The patent applies partial action by performing clustering only on identified motion periods that meet specific criteria (angular velocity ranges, duration thresholds) rather than processing all collected data. This selective approach maintains high identification accuracy for repetitive motions while minimizing unnecessary processing time
Data Source
AI summary
A detecting method for repetitive motion includes receiving a plurality of accelerated velocities, receiving a plurality of angular velocities, determining a plurality of stationary points according to the accelerated velocities, determining a plurality of motion periods according to the stationary points, wherein the motion periods separately correspond to different sets of the angular velocities, and clustering the motion periods according to differences among the different sets of angular velocities corresponding to the motion periods.